Jason Merrill [Thu, 3 Dec 2020 21:38:19 +0000 (16:38 -0500)]
c++: Fix bootstrap on 32-bit hosts [PR91828]
Using the releasing_vec op[] with an int index was breaking on 32-bit hosts
because of ambiguity with the built-in operator and the conversion
function. Since the built-in operator has a ptrdiff_t, this was fine on
64-bit targets where ptrdiff_t is larger than int, but broke on 32-bit
targets where it's the same as int, making the conversion for that argument
better than the member function. Fixed by changing the member function to
also use ptrdiff_t for the index.
gcc/cp/ChangeLog:
* cp-tree.h (releasing_vec::operator[]): Change parameter type to
ptrdiff_t.
Martin Sebor [Thu, 3 Dec 2020 22:41:25 +0000 (15:41 -0700)]
Add support for detecting mismatched allocation/deallocation calls.
PR c++/90629 - Support for -Wmismatched-new-delete
PR middle-end/94527 - Add an __attribute__ that marks a function as freeing an object
gcc/ChangeLog:
PR c++/90629
PR middle-end/94527
* builtins.c (access_ref::access_ref): Initialize new member.
(compute_objsize): Use access_ref::deref. Handle simple pointer
assignment.
(expand_builtin): Remove handling of the free built-in.
(call_dealloc_argno): Same.
(find_assignment_location): New function.
(fndecl_alloc_p): Same.
(gimple_call_alloc_p): Same.
(call_dealloc_p): Same.
(matching_alloc_calls_p): Same.
(warn_dealloc_offset): Same.
(maybe_emit_free_warning): Same.
* builtins.h (struct access_ref): Declare new member.
(maybe_emit_free_warning): Make extern. Make use of access_ref.
Handle -Wmismatched-new-delete.
* calls.c (initialize_argument_information): Call
maybe_emit_free_warning.
* doc/extend.texi (attribute malloc): Update.
* doc/invoke.texi (-Wfree-nonheap-object): Expand documentation.
(-Wmismatched-new-delete): Document new option.
(-Wmismatched-dealloc): Document new option.
gcc/c-family/ChangeLog:
PR c++/90629
PR middle-end/94527
* c-attribs.c (handle_dealloc_attribute): New function.
(handle_malloc_attribute): Handle argument forms of attribute.
* c.opt (-Wmismatched-dealloc): New option.
(-Wmismatched-new-delete): New option.
gcc/testsuite/ChangeLog:
PR c++/90629
PR middle-end/94527
* g++.dg/asan/asan_test.cc: Fix a bug.
* g++.dg/warn/delete-array-1.C: Add expected warning.
* g++.old-deja/g++.other/delete2.C: Add expected warning.
* g++.dg/warn/Wfree-nonheap-object-2.C: New test.
* g++.dg/warn/Wfree-nonheap-object.C: New test.
* g++.dg/warn/Wmismatched-new-delete.C: New test.
* g++.dg/warn/Wmismatched-dealloc-2.C: New test.
* g++.dg/warn/Wmismatched-dealloc.C: New test.
* gcc.dg/Wmismatched-dealloc.c: New test.
* gcc.dg/analyzer/malloc-1.c: Prune out expected warning.
* gcc.dg/attr-malloc.c: New test.
* gcc.dg/free-1.c: Adjust text of expected warning.
* gcc.dg/free-2.c: Same.
* gcc.dg/torture/pr71816.c: Prune out expected warning.
* gcc.dg/tree-ssa/pr19831-2.c: Add an expected warning.
* gcc.dg/Wfree-nonheap-object-2.c: New test.
* gcc.dg/Wfree-nonheap-object-3.c: New test.
* gcc.dg/Wfree-nonheap-object.c: New test.
libstdc++-v3/ChangeLog:
* testsuite/ext/vstring/modifiers/clear/56166.cc: Suppress a false
positive warning.
Nathan Sidwell [Thu, 3 Dec 2020 20:38:15 +0000 (12:38 -0800)]
c++: Exported using decls
With modules we need to record whethe a (namespace-scope) using decl
is exporting the named entities. Record this on the OVERLOAD marking
the used decl.
gcc/cp/
* cp-tree.h (OVL_EXPORT): New.
(class ovl_iterator): Add get_using, exporting_p.
* tree.c (ovl_insert): Extend using_or_hidden meaning to include
an exported using.
Nathan Sidwell [Thu, 3 Dec 2020 19:12:08 +0000 (11:12 -0800)]
c++: uninstantiated template friends
template friends need to be recognized by module streaming and
associated with the befriending class. but their context is that of
the friend (a namespace or other class). This adds a flag to mark
such templates, and uses their DECL_CHAIN to point at the befriender.
gcc/cp
* cp-tree.h (DECL_UNINSTANTIATED_TEMPLATE_FRIEND): New.
* pt.c (push_template_decl): Set it.
(tsubst_friend_function): Clear it.
PowerPC: PR libgcc/97543 and libgcc/97643, fix long double issues
If you use a compiler with long double defaulting to 64-bit instead of 128-bit
with IBM extended double, you get linker warnings about mis-matches in the gnu
attributes for long double (PR libgcc/97543). Even if the compiler is
configured to have long double be 64 bit as the default with the configuration
option '--without-long-double-128' you get the warnings.
You also get the same issues if you use a compiler with long double defaulting
to IEEE 128-bit instead of IBM extended double (PR libgcc/97643).
The issue is the way libgcc.a/libgcc.so is built. Right now when building
libgcc under Linux, the long double size is set to 128-bits when building
libgcc. However, the gnu attributes are set, leading to the warnings.
One feature of the current GNU attribute implementation is if you have a shared
library (such as libgcc_s.so), the GNU attributes for the shared library is an
inclusive OR of all of the objects within the library. This means if any
object file that uses the -mlong-double-128 option and uses long double, the GNU
attributes for the library will indicate that it uses 128-bit IBM long
doubles. If you have a static library, you will get the warning only if you
actually reference an object file with the attribute set.
This patch does two things:
1) All of the object files that support IBM 128-bit long doubles
explicitly set the ABI to IBM extended double.
2) I turned off GNU attributes for building the shared library or for
building the IBM 128-bit long double support.
libgcc/
2020-12-03 Michael Meissner <meissner@linux.ibm.com>
PR libgcc/97543
PR libgcc/97643
* config/rs6000/t-linux (IBM128_STATIC_OBJS): New make variable.
(IBM128_SHARED_OBJS): New make variable.
(IBM128_OBJS): New make variable. Set all objects to use the
explicit IBM format, and disable gnu attributes.
(IBM128_CFLAGS): New make variable.
(gcc_s_compile): Add -mno-gnu-attribute to all shared library
modules.
Alexandre Oliva [Thu, 3 Dec 2020 19:20:28 +0000 (16:20 -0300)]
fix __builtin___clear_cache overrider fallout
Machines that had CLEAR_CACHE_INSN and that would thus issue calls to
__clear_cache with the default call expander, would fail on languages
that did not set up the __clear_cache builtin. This patch arranges
for all languages to set up this builtin.
Machines or multilibs that had ptr_mode != Pmode, such as aarch64 with
-mabi=ilp32, would fail the RTL mode test of the arguments passed to
__clear_cache, because we'd insist on ptr_mode. This patch arranges
for Pmode to be accepted as well.
for gcc/ChangeLog
* tree.c (build_common_builtin_nodes): Declare
__builtin___clear_cache for all languages.
* builtins.c (maybe_emit_call_builtin___clear_cache): Accept
Pmode arguments.
This doesn't define a new _GLIBCXX_HAVE_BUILTIN_SOURCE_LOCATION macro.
because using __has_builtin(__builtin_source_location) is sufficient.
Currently only GCC supports it, but if/when Clang and Intel add it the
__has_builtin check should for them too.
Co-authored-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
* doc/doxygen/user.cfg.in (INPUT): Add <source_location>.
* include/Makefile.am: Add <source_location>.
* include/Makefile.in: Regenerate.
* include/std/version (__cpp_lib_source_location): Define.
* include/std/source_location: New file.
* testsuite/18_support/source_location/1.cc: New test.
* testsuite/18_support/source_location/consteval.cc: New test.
* testsuite/18_support/source_location/srcloc.h: New test.
* testsuite/18_support/source_location/version.cc: New test.
Jonathan Wakely [Thu, 3 Dec 2020 19:17:13 +0000 (19:17 +0000)]
libstdc++: Add std::bit_cast for C++20 [PR 93121]
Thanks to Jakub's addition of the built-in, we can add this to the
library now. The compiler tests for the built-in are quite extensive,
including verifying the constraints, so this only adds minimal tests to
the library testsuite.
This doesn't add a new _GLIBCXX_HAVE_BUILTIN_BIT_CAST because using
__has_builtin(__builtin_bit_cast) works for GCC and versions of Clang
that provide the built-in.
libstdc++-v3/ChangeLog:
PR libstdc++/93121
* include/std/bit (__cpp_lib_bit_cast, bit_cast): Define.
* include/std/version (__cpp_lib_bit_cast): Define.
* testsuite/26_numerics/bit/bit.cast/bit_cast.cc: New test.
* testsuite/26_numerics/bit/bit.cast/version.cc: New test.
Jonathan Wakely [Thu, 3 Dec 2020 17:08:01 +0000 (17:08 +0000)]
libstdc++: Disable std::array assertions for C++11 constexpr
The recent changes to add assertions to std::array broke the functions
that need to be constexpr in C++11, because of the restrictive rules for
constexpr functions in C++11.
This simply disables the assertions for C++11 mode, so the functions can
be constexpr again.
libstdc++-v3/ChangeLog:
* include/std/array (array::operator[](size_t) const, array::front() const)
(array::back() const) [__cplusplus == 201103]: Disable
assertions.
* testsuite/23_containers/array/element_access/constexpr_element_access.cc:
Check for correct values.
* testsuite/23_containers/array/tuple_interface/get_neg.cc:
Adjust dg-error line numbers.
* testsuite/23_containers/array/debug/constexpr_c++11.cc: New test.
Nathan Sidwell [Thu, 3 Dec 2020 16:58:44 +0000 (08:58 -0800)]
c++: templatey type creation
This patch makes a couple of type-creation routines available to
modules. That needs to create unbound template parms, and canonical
template parms.
gcc/cp/
* cp-tree.h (make_unbound_class_template_raw): Declare.
(canonical_type_parameter): Declare.
* decl.c (make_unbound_class_template_raw): Break out of ...
(make_unboud_class_template): ... here. Call it.
* pt.c (canonical_type_parameter): Externalize. Refactor & set
structural_equality for type parms.
Uros Bizjak [Thu, 3 Dec 2020 16:49:42 +0000 (17:49 +0100)]
i386: Fix up ix86_md_asm_adjust for TImode [PR98086]
ix86_md_asm_adjust assumes that dest_mode can be only [QHSD]Imode
and nothing else. The patch rewrites zero-extension part to use
convert_to_mode to handle TImode and hypothetically even wider modes.
2020-12-03 Uroš Bizjak <ubizjak@gmail.com>
Jakub Jelinek <jakub@redhat.com>
gcc/
PR target/98086
* config/i386/i386.c (ix86_md_asm_adjustmd): Rewrite
zero-extension part to use convert_to_mode.
gcc/testsuite/
PR target/98086
* gcc.target/i386/pr98086.c: New test.
compiler: defer to middle-end for complex division
Go used to use slightly different semantics than C99 for complex division,
so we used runtime routines to handle the different. The gc compiler
has changes its behavior to match C99, so changes ours as well.
Andreas Krebbel [Thu, 3 Dec 2020 16:04:11 +0000 (17:04 +0100)]
IBM Z: Fix mode in probe_stack pattern
The probe pattern uses Pmode but the middle-end wants to emit a
word_mode probe check. This - as usual - breaks on Z with -m31
-mzarch were word_mode doesn't match Pmode.
gcc/ChangeLog:
* config/s390/s390.md ("@probe_stack2<mode>"): Change mode
iterator to W.
Nathan Sidwell [Thu, 3 Dec 2020 15:42:09 +0000 (07:42 -0800)]
c++: Fix array type dependency [PR 98107]
I'd missed some paths through build_cplus_array_type, plus, some
arrays come via the C-type builder. This propagates dependency in
more places and asserts that in the cases where TYPE_DEPENDENT_P_VALID
is unset, the type is non-dependent.
PR c++/98107
gcc/cp/
* tree.c (build_cplus_array_type): Mark dependency of new variant.
(cp_build_qualified_type_real, strip_typedefs): Assert
TYPE_DEPENDENT_P_VALID, or not a dependent type.
As the testcase shows, folding svundef*() at the gimple level
has the unfortunate side-effect of introducing -Wuninitialized
or -Wmaybe-uninitialized warnings. We don't have a testcase
that relies on the fold, so the easiest fix seems to be to
remove it.
Eric Botcazou [Thu, 3 Dec 2020 14:54:14 +0000 (15:54 +0100)]
Fix PR middle-end/98099
this replaces the ICE by a sorry message for the use of reverse scalar
storage order with 128-bit decimal floating-point type on 32-bit targets.
gcc/ChangeLog:
PR middle-end/98099
* expmed.c (flip_storage_order): In the case of a non-integer mode,
sorry out if the integer mode to be used instead is not supported.
gcc/testsuite/ChangeLog:
* gcc.dg/pr98099.c: New test.
Eric Botcazou [Thu, 3 Dec 2020 14:52:15 +0000 (15:52 +0100)]
Fix PR middle-end/98082
this fixes an ICE introduced by the fix for PR middle-end/97078 where
use_register_for_decl was changed to return true at -O0 for a parameter
of a thunk. It turns out that we need to do the same for a result in
this case.
gcc/ChangeLog:
PR middle-end/98082
* function.c (use_register_for_decl): Also return true for a result
if cfun->tail_call_marked is true.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/pr98082.C: New test.
Jakub Jelinek [Thu, 3 Dec 2020 14:46:54 +0000 (15:46 +0100)]
c++: Add __builtin_bit_cast to implement std::bit_cast [PR93121]
The following patch adds __builtin_bit_cast builtin, similarly to
clang or MSVC which implement std::bit_cast using such an builtin too.
It checks the various std::bit_cast requirements, when not constexpr
evaluated acts pretty much like VIEW_CONVERT_EXPR of the source argument
to the destination type and the hardest part is obviously the constexpr
evaluation.
I've left out PDP11 handling of those, couldn't figure out how exactly are
bitfields laid out there
2020-12-03 Jakub Jelinek <jakub@redhat.com>
PR libstdc++/93121
* fold-const.h (native_encode_initializer): Add mask argument
defaulted to nullptr.
(find_bitfield_repr_type): Declare.
(native_interpret_aggregate): Declare.
* fold-const.c (find_bitfield_repr_type): New function.
(native_encode_initializer): Add mask argument and support for
filling it. Handle also some bitfields without integral
DECL_BIT_FIELD_REPRESENTATIVE.
(native_interpret_aggregate): New function.
* gimple-fold.h (clear_type_padding_in_mask): Declare.
* gimple-fold.c (struct clear_padding_struct): Add clear_in_mask
member.
(clear_padding_flush): Handle buf->clear_in_mask.
(clear_padding_union): Copy clear_in_mask. Don't error if
buf->clear_in_mask is set.
(clear_padding_type): Don't error if buf->clear_in_mask is set.
(clear_type_padding_in_mask): New function.
(gimple_fold_builtin_clear_padding): Set buf.clear_in_mask to false.
* doc/extend.texi (__builtin_bit_cast): Document.
* g++.dg/cpp2a/bit-cast1.C: New test.
* g++.dg/cpp2a/bit-cast2.C: New test.
* g++.dg/cpp2a/bit-cast3.C: New test.
* g++.dg/cpp2a/bit-cast4.C: New test.
* g++.dg/cpp2a/bit-cast5.C: New test.
Jakub Jelinek [Thu, 3 Dec 2020 14:32:31 +0000 (15:32 +0100)]
c++: consteval-defarg1.C test variant for templates
We weren't recognizing a default argument for a consteval member function as
being in immediate function context because there was no function parameter
scope to look at.
The following testcase is an attempt to test it with templates, both
non-dependent and dependent consteval calls in both function and class
templates, and with r11-5694 it now passes.
tree-ssa-threadedge.c (record_temporary_equivalences_from_stmts_at_dest): Do not allow __builtin_constant_p.
This is the same as commit 70a62009181f ("tree-ssa-threadbackward.c
(profitable_jump_thread_path): Do not allow __builtin_constant_p."), but
for the old forward threader.
gcc/ChangeLog:
2020-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
* tree-ssa-threadedge.c (record_temporary_equivalences_from_stmts_at_dest):
Do not allow __builtin_constant_p on a threading path.
Kito Cheng [Wed, 25 Nov 2020 09:30:29 +0000 (17:30 +0800)]
RISC-V: Canonicalize --with-arch
- We would like to canonicalize the arch string for --with-arch for
easier handling multilib, so split canonicalization part to a stand
along script to shared the logic.
gcc/ChangeLog:
* config/riscv/multilib-generator (arch_canonicalize): Move
code to arch-canonicalize, and call that script to canonicalize arch
string.
(canonical_order): Move code to arch-canonicalize.
(LONG_EXT_PREFIXES): Ditto.
(IMPLIED_EXT): Ditto.
* config/riscv/arch-canonicalize: New.
* config.gcc (riscv*-*-*): Canonicalize --with-arch.
New +flagm (Condition flag manipulation) feature option for -march command line
option.
Please note that FLAGM stays a Armv8.4-A feature but now can be
assigned to other architectures or CPUs.
gcc/ChangeLog:
* config/aarch64/aarch64-option-extensions.def
(AARCH64_OPT_EXTENSION): New +flagm option in -march for AArch64.
* config/aarch64/aarch64.h (AARCH64_FL_FLAGM): Add new flagm extension bit
mask.
(AARCH64_FL_FOR_ARCH8_4): Add flagm to Armv8.4-A.
* doc/invoke.texi: Update docs with +flagm.
liuhongt [Mon, 30 Nov 2020 05:27:16 +0000 (13:27 +0800)]
Optimize vpsubusw compared to 0 into vpcmpleuw or vpcmpnleuw [PR96906]
For signed comparisons, it handles cases that are eq or neq to 0.
For unsigned comparisons, it additionaly handles cases that are le or
gt to 0(equivilent to eq or neq to 0). Transform case eq to leu,
case neq to gtu.
.i.e. for -mavx512bw -mavx512vl transform eq case code from
liuhongt [Tue, 3 Nov 2020 09:26:43 +0000 (17:26 +0800)]
Fix incorrect replacement of vmovdqu32 with vpblendd which can cause fault.
gcc/ChangeLog:
PR target/97642
* config/i386/i386-expand.c
(ix86_expand_special_args_builtin): Don't move all-ones mask
operands into register.
* config/i386/sse.md (UNSPEC_MASKLOAD): New unspec.
(*<avx512>_load<mode>_mask): New define_insns for masked load
instructions.
(<avx512>_load<mode>_mask): Changed to define_expands which
specifically handle memory or all-ones mask operands.
(<avx512>_blendm<mode>): Changed to define_insns which are same
as original <avx512>_load<mode>_mask with adjustment of
operands order.
(*<avx512>_load<mode>): New define_insn_and_split which is
used to optimize for masked load with all one mask.
gcc/testsuite/ChangeLog:
* gcc.target/i386/avx512bw-vmovdqu16-1.c: Adjust testcase to
make sure only masked load instruction is generated.
* gcc.target/i386/avx512bw-vmovdqu8-1.c: Ditto.
* gcc.target/i386/avx512f-vmovapd-1.c: Ditto.
* gcc.target/i386/avx512f-vmovaps-1.c: Ditto.
* gcc.target/i386/avx512f-vmovdqa32-1.c: Ditto.
* gcc.target/i386/avx512f-vmovdqa64-1.c: Ditto.
* gcc.target/i386/avx512vl-vmovapd-1.c: Ditto.
* gcc.target/i386/avx512vl-vmovaps-1.c: Ditto.
* gcc.target/i386/avx512vl-vmovdqa32-1.c: Ditto.
* gcc.target/i386/avx512vl-vmovdqa64-1.c: Ditto.
* gcc.target/i386/pr97642-1.c: New test.
* gcc.target/i386/pr97642-2.c: New test.
Jason Merrill [Wed, 2 Dec 2020 22:11:48 +0000 (17:11 -0500)]
c++: Push parms when late parsing default args
In this testcase we weren't catching the error in A::f because the parameter
'I' wasn't in scope, so the default argument for 'b' found the global
typedef I. Fixed by pushing the parms before parsing. This is a bit
complicated because pushdecl clears DECL_CHAIN; do_push_parm_decls deals
with this by nreversing first, but that doesn't work here because we only
want to push them one at a time; if we pushed all of them before parsing,
we'd wrongly reject A::g.
gcc/cp/ChangeLog:
* parser.c (cp_parser_primary_expression): Distinguish
parms from vars in error.
(cp_parser_late_parsing_default_args): Pushdecl parms
as we go.
Jason Merrill [Wed, 2 Dec 2020 19:35:50 +0000 (14:35 -0500)]
c++: Fix late-parsed default arg context
Jakub noticed that we weren't recognizing a default argument for a consteval
member function as being in immediate function context because there was no
function parameter scope to look at.
Note that this patch doesn't actually push the parameters into the scope,
that happens in a separate commit.
Alexandre Oliva [Thu, 3 Dec 2020 01:10:32 +0000 (22:10 -0300)]
introduce overridable clear_cache emitter
This patch introduces maybe_emit_call_builtin___clear_cache for the
builtin expander machinery and the trampoline initializers to use to
clear the instruction cache, removing a source of inconsistencies and
subtle errors in low-level machinery.
I've adjusted all trampoline_init implementations that used to issue
explicit calls to __clear_cache or similar to use this new primitive.
Specifically on vxworks targets, we needed to drop the __clear_cache
symbol in libgcc, for reasons related with linking that I didn't need
to understand, and we wanted to call cacheTextUpdate directly, despite
the different calling conventions: the second argument is a length
rather than the end address.
So I introduced a target hook to enable target OS-level overriding of
builtin __clear_cache call emission, retaining nearly (*) the same
logic to govern the decision on whether to emit a call (or nothing, or
a machine-dependent insn) but enabling a call to a target
system-defined function with different calling conventions to be
issued, without having to modify .md files of the various
architectures supported by the target system to introduce or modify
clear_cache insns.
(*) I write "nearly" mainly because, when not optimizing, we'd issue a
call regardless, but since the call may now be overridden, I added it
to the set of builtins that are not directly turned into calls when
not optimizing, following the normal expansion path instead. It
wouldn't be hard to skip the emission of cache-clearing insns when not
optimizing, but it didn't seem very important, especially for the new
uses from trampoline init.
Another difference that might be relevant is that now we expand
the begin and end arguments unconditionally. This might make a
difference if they have side effects. That's prettty much impossible
at expand time, but I thought I'd mention it.
I have NOT modified targets that did not issue cache-clearing calls in
trampoline init to use the new clear_cache-calling infrastructure even
if it would expand to nothing. I have considered doing so, to have
__builtin___clear_cache and trampoline init call cacheTextUpdate on
all vxworks targets, but decided not to, since on targets that don't
do any cache clearing, cacheTextUpdate ought to be a no-op, even
though rs6000 seems to use icbi and dcbf instructions in the function
called to initialize a trampoline, but AFAICT not in the __clear_cache
builtin. Hopefully target maintainers will have a look and take
advantage of this new piece of infrastructure to remove such
(apparent?) inconsistencies. Not rs6000 and other that call asm-coded
trampoline setup instructions, for sure, but they might wish to
introduce a CLEAR_INSN_CACHE macro or a clear_cache expander if they
don't have one.
for gcc/ChangeLog
* builtins.c (default_emit_call_builtin___clear_cache): New.
(maybe_emit_call_builtin___clear_cache): New.
(expand_builtin___clear_cache): Split into the above.
(expand_builtin): Do not issue clear_cache call any more.
* builtins.h (maybe_emit_call_builtin___clear_cache): Declare.
* config/aarch64/aarch64.c (aarch64_trampoline_init): Use
maybe_emit_call_builtin___clear_cache.
* config/arc/arc.c (arc_trampoline_init): Likewise.
* config/arm/arm.c (arm_trampoline_init): Likewise.
* config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
* config/csky/csky.c (csky_trampoline_init): Likewise.
* config/m68k/linux.h (FInALIZE_TRAMPOLINE): Likewise.
* config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
* config/tilepro/tilepro.c (tilepro_trampoline_init): Ditto.
* config/vxworks.c: Include rtl.h, memmodel.h, and optabs.h.
(vxworks_emit_call_builtin___clear_cache): New.
* config/vxworks.h (CLEAR_INSN_CACHE): Drop.
(TARGET_EMIT_CALL_BUILTIN___CLEAR_CACHE): Define.
* target.def (trampoline_init): In the documentation, refer to
maybe_emit_call_builtin___clear_cache.
(emit_call_builtin___clear_cache): New.
* doc/tm.texi.in: Add new hook point.
(CLEAR_CACHE_INSN): Remove duplicate 'both'.
* doc/tm.texi: Rebuilt.
* targhooks.h (default_meit_call_builtin___clear_cache):
Declare.
* tree.h (BUILTIN_ASM_NAME_PTR): New.
Alexandre Oliva [Thu, 3 Dec 2020 01:10:26 +0000 (22:10 -0300)]
options.exp: unsupport tests that depend on missing language
There's a help.exp test that checks that the help message for
-Wabsolute-value mentions it's available in C and ObjC, when compiling
a C++ program.
However, if GCC is built with the C++ language disabled, the
.cc file is compiled as C, and the message [available in C...] becomes
[disabled] instead, because that's the default for the flag in C.
I suppose it might also be possible to disable the C language, and
then the multitude of help.exp tests that name c as the source
language will fail.
This patch avoids these fails: it detects the message "compiler not
installed" in the compiler output, and bails out as "unsupported".
for gcc/testsuite/ChangeLog
* lib/options.exp (check_for_options_with_filter): Detect
unavailable compiler for the selected language, and bail out
as unsupported.
tree-ssa-threadbackward.c (profitable_jump_thread_path): Do not allow __builtin_constant_p.
Linux Kernel (specifically, drivers/leds/trigger/ledtrig-cpu.c) build
with GCC 10 fails on s390 with "impossible constraint".
Explanation by Jeff Law:
```
So what we have is a b_c_p at the start of an if-else chain. Subsequent
tests on the "true" arm of the the b_c_p test may throw us off the
constant path (because the constants are out of range). Once all the
tests are passed (it's constant and the constant is in range) the true
arm's terminal block has a special asm that requires a constant
argument. In the case where we get to the terminal block on the true
arm, the argument to the b_c_p is used as the constant argument to the
special asm.
At first glace jump threading seems to be doing the right thing. Except
that we end up with two paths to that terminal block with the special
asm, one for each of the two constant arguments to the b_c_p call.
Naturally since that same value is used in the asm, we have to introduce
a PHI to select between them at the head of the terminal block. Now
the argument in the asm is no longer constant and boom we fail.
```
Fix by disallowing __builtin_constant_p on threading paths.
gcc/ChangeLog:
2020-06-03 Ilya Leoshkevich <iii@linux.ibm.com>
* tree-ssa-threadbackward.c (thread_jumps::profitable_jump_thread_path):
Do not allow __builtin_constant_p on a threading path.
gcc/testsuite/ChangeLog:
2020-06-03 Ilya Leoshkevich <iii@linux.ibm.com>
* gcc.target/s390/builtin-constant-p-threading.c: New test.
Peter Bergner [Thu, 3 Dec 2020 00:12:59 +0000 (18:12 -0600)]
c++: Treat OPAQUE_TYPE types as an aggregate type [PR97947]
MODE_OPAQUE and the associated OPAQUE_TYPE were added to stop the optimizers
from knowing how the bits in a variable with an opaque type are laid out.
This makes them a kind of pseudo aggregate type and we need to treat them
as such when we process the INIT initializer for variables with an
opaque type.
2020-12-02 Peter Bergner <bergner@linux.ibm.com>
gcc/cp/
PR c++/97947
* typeck2.c (digest_init_r): Handle OPAQUE_TYPE as an aggregate type.
gcc/testsuite/
PR c++/97947
* g++.target/powerpc/pr97947.C: New test.
Jakub Jelinek [Wed, 2 Dec 2020 23:29:46 +0000 (00:29 +0100)]
dwarf2out: Fix up add_scalar_info not to create invalid DWARF
As discussed in https://sourceware.org/bugzilla/show_bug.cgi?id=26987 ,
for very large bounds (which don't fit into HOST_WIDE_INT) GCC emits invalid
DWARF.
In DWARF2, DW_AT_{lower,upper}_bound were constant reference class.
In DWARF3 they are block constant reference and the
Static and Dynamic Properties of Types
chapter says:
"For a block, the value is interpreted as a DWARF expression; evaluation of the expression
yields the value of the attribute."
In DWARF4/5 they are constant exprloc reference class.
Now, for add_AT_wide we use DW_FORM_data16 (valid in constant class)
when -gdwarf-5, but otherwise just use DW_FORM_block1, which is not constant
class, but block.
For DWARF3 this means emitting clearly invalid DWARF, because the
DW_FORM_block1 should contain a DWARF expression, not random bytes
containing the constant directly.
For DWARF2/DWARF4/5 it could be considered a GNU extension, but a very badly
designed one when it means something different in DWARF3.
The following patch uses add_AT_wide only if we know we'll be using
DW_FORM_data16, and otherwise wastes 2 extra bytes and emits in there
DW_OP_implicit_value <size> before the constant.
2020-12-03 Jakub Jelinek <jakub@redhat.com>
* dwarf2out.c (add_scalar_info): Only use add_AT_wide for 128-bit
constants and only in dwarf-5 or later, where DW_FORM_data16 is
available. Otherwise use DW_FORM_block*/DW_FORM_exprloc with
DW_OP_implicit_value to describe the constant.
Jakub Jelinek [Wed, 2 Dec 2020 23:25:51 +0000 (00:25 +0100)]
c++: Implement LWG3396 Clarify point of reference for source_location::current() [PR80780, PR93093]
While std::source_location::current () is static consteval source_location
current() noexcept; in the standard, it also says with LWG3396:
"Any call to current that appears as a default member initializer
([class.mem]), or as a subexpression thereof, should correspond to the
location of the constructor definition or aggregate initialization that uses
the default member initializer. Any call to current that appears as a
default argument ([dcl.fct.default]), or as a subexpression thereof, should
correspond to the location of the invocation of the function that uses the
default argument ([expr.call])."
so it must work as compiler magic rather than normal immediate functions,
in particular we need to defer its evaluation when parsing default arguments
or nsdmis.
This patch actually defers evaluation of all the calls to
std::source_location::current () until genericization (or constant expression
evaluation when called from constant expression contexts).
I had to change constexpr.c too so that it temporarily adjusts
current_function_decl from the constexpr evaluation context, but we do the
same already from __builtin_FUNCTION ().
2020-12-03 Jakub Jelinek <jakub@redhat.com>
PR c++/80780
PR c++/93093
* cp-tree.h (source_location_current_p): Declare.
* tree.c (source_location_current_p): New function.
* call.c (immediate_invocation_p): New function.
(build_over_call): Use it to resolve LWG3396.
* constexpr.c (cxx_eval_builtin_function_call): Temporarily set
current_function_decl from ctx->call->fundef->decl if any.
* cp-gimplify.c (cp_genericize_r) <case CALL_EXPR>: Fold calls
to immediate function std::source_location::current ().
* g++.dg/cpp2a/srcloc15.C: New test.
* g++.dg/cpp2a/srcloc16.C: New test.
* g++.dg/cpp2a/srcloc17.C: New test.
* g++.dg/cpp2a/srcloc18.C: New test.
Jonathan Wakely [Wed, 2 Dec 2020 21:39:08 +0000 (21:39 +0000)]
libstdc++: Fix std::any pretty printer [PR 68735]
This fixes errors seen on powerpc64 (big endian only) due to the
printers for std::any and std::experimental::any being unable to find
the manager function.
libstdc++-v3/ChangeLog:
PR libstdc++/65480
PR libstdc++/68735
* python/libstdcxx/v6/printers.py (function_pointer_to_name):
New helper function to get the name of a function from its
address.
(StdExpAnyPrinter.__init__): Use it.
Jason Merrill [Tue, 24 Nov 2020 23:21:38 +0000 (18:21 -0500)]
c++: Improve init handling
While looking at another issue I noticed that in a template we were failing
to find the INIT_EXPR we were looking for, and so ended up doing redundant
processing. No testcase, as the redundant processing ended up getting the
right result.
gcc/cp/ChangeLog:
* decl.c (check_initializer): Also look through STMT_EXPR
and BIND_EXPR.
Nathan Sidwell [Wed, 2 Dec 2020 20:53:23 +0000 (12:53 -0800)]
c++: typename_type structural comparison
For modules we need to compare structurally all the way down. This
means inhibiting typename_type resolution, independent of comparing
specializations.
Marek Polacek [Tue, 1 Dec 2020 15:39:08 +0000 (10:39 -0500)]
c++: Fix ICE with inline variable in template [PR97975]
In this test, we have
static inline const int c = b;
in a class template, and we call store_init_value as usual. There, the
value is
IMPLICIT_CONV_EXPR<const float>(b)
which is is_nondependent_static_init_expression but isn't
is_nondependent_constant_expression (they only differ in STRICT).
We call fold_non_dependent_expr, but that just returns the expression
because it only instantiates is_nondependent_constant_expression
expressions. Since we're not checking the initializer of a constexpr
variable, we go on to call maybe_constant_init, whereupon we crash
because it tries to evaluate all is_nondependent_static_init_expression
expressions, which our value is, but it still contains a template code.
I think the fix is to call fold_non_dependent_init instead of
maybe_constant_init, and only call fold_non_dependent_expr on the
"this is a constexpr variable" path so as to avoid instantiating twice
in a row. Outside a template this should also avoid evaluating the
value twice.
gcc/cp/ChangeLog:
PR c++/97975
* constexpr.c (fold_non_dependent_init): Add a tree parameter.
Use it.
* cp-tree.h (fold_non_dependent_init): Add a tree parameter with
a default value.
* typeck2.c (store_init_value): Call fold_non_dependent_expr
only when checking the initializer for constexpr variables.
Call fold_non_dependent_init instead of maybe_constant_init.
gcc/testsuite/ChangeLog:
PR c++/97975
* g++.dg/cpp1z/inline-var8.C: New test.
Marek Polacek [Mon, 30 Nov 2020 19:53:24 +0000 (14:53 -0500)]
c++: Fix tsubst ICE with invalid code [PR97993, PR97187]
I had a strong sense of deja vu when looking into this, and no wonder,
since this is almost identical to c++/95728.
Since r11-423 tsubst_copy_and_build/TREE_LIST uses tsubst_tree_list
instead of open coding it. While the latter could return an error
node wrapped in a TREE_LIST, the former can return a naked error node.
That broke in tsubst_copy_and_build/NEW_EXPR, because we were accessing
TREE_VALUE of an error node.
gcc/cp/ChangeLog:
PR c++/97187
PR c++/97993
* pt.c (tsubst_copy_and_build) <case NEW_EXPR>: Return error_mark_node
if init is erroneous.
gcc/testsuite/ChangeLog:
PR c++/97187
PR c++/97993
* g++.dg/eh/crash2.C: New test.
* g++.dg/template/crash132.C: New test.
Ilya Leoshkevich [Thu, 26 Nov 2020 01:16:17 +0000 (02:16 +0100)]
IBM Z: Use llihf and oilf to load large immediates into GPRs
Currently GCC loads large immediates into GPRs from the literal pool,
which is not as efficient as loading two halves with llihf and oilf.
gcc/ChangeLog:
2020-11-30 Ilya Leoshkevich <iii@linux.ibm.com>
* config/s390/s390-protos.h (s390_const_int_pool_entry_p): New
function.
* config/s390/s390.c (s390_const_int_pool_entry_p): New
function.
* config/s390/s390.md: Add define_peephole2 that produces llihf
and oilf.
gcc/testsuite/ChangeLog:
2020-11-30 Ilya Leoshkevich <iii@linux.ibm.com>
* gcc.target/s390/load-imm64-1.c: New test.
* gcc.target/s390/load-imm64-2.c: New test.
Jonathan Wakely [Wed, 2 Dec 2020 16:37:56 +0000 (16:37 +0000)]
libstdc++: Use libatomic for tests on all 32-bit powerpc targets
In addition to the existing powerpc targets, powerpc64 needs libatomic
for 64-bit atomics when testing the 32-bit multilib with -m32. Adjust
the existing target checks to match all 32-bit powerpc targets, but not
64-bit ones.
libstdc++-v3/ChangeLog:
* testsuite/lib/dg-options.exp (add_options_for_libatomic):
Replace powerpc-ibm-aix* and powerpc*-*-darwin* with check for
powerpc && ilp32.
Nathan Sidwell [Wed, 2 Dec 2020 16:27:53 +0000 (08:27 -0800)]
c++: RTTI accessors for modules
The module machinery needs to serialize tinfo types and vars by
meaning, not literally. This adds the necessary pieces to rtti.
gcc/cp/
* cp-tree.h (DECL_TINFO_P): Also for TYPE_DECLs.
(get_tinfo_decl_direct): Declare.
(get_pseudo_tinfo_index, get_pseudo_tinfo_type): Declare.
* rtti.c (get_tinfo_decl_direct): Externalize.
(get_tinfo_desc): Set DECL_TINFO_P on the typedef.
(get_pseudo_tinfo_index, get_pseudo_tinfo_type): New.
Ian Lance Taylor [Fri, 27 Nov 2020 04:13:16 +0000 (20:13 -0800)]
compiler: reword "declared and not used" error message
This is a gofrontend copy of https://golang.org/cl/203282.
From the CL 203282 description:
"declared and not used" is technically correct, but might confuse
the user. Switching "and" to "but" will hopefully create the
contrast for the users: they did one thing (declaration), but
not the other --- actually using the variable.
Nathan Sidwell [Wed, 2 Dec 2020 15:35:23 +0000 (07:35 -0800)]
c++: Extend build_array_type API
The modules machinery needs to construct array types, and that wanted
to determine type dependency. That doesn't work during loading, but
the module loader can stream that fact too and tell the array builder.
Thus this extends the API to allow a caller to specify the dependency
explicitly. The call in cp_build_qualified_type_real is unreachable
during the loading, so that one's ok as is.
gcc/cp/
* cp-tree.h (build_cplus_array_type): Add defaulted DEP parm.
* tree.c (set_array_type_common): Add DEP parm.
(build_cplus_array_type): Add DEP parm, determine dependency if
needed.
(cp_build_qualified_type_real): Adjust array-building call.
(strip_typedefs): Likewise.
Scott Snyder [Wed, 2 Dec 2020 14:42:56 +0000 (15:42 +0100)]
vec.h: Fix GCC build with -std=gnu++20 [PR98059]
Apparently vec.h doesn't build with -std=c++20/gnu++20, since the
DR2237 r11-532 change.
template <typename T>
class auto_delete_vec
{
private:
auto_vec_delete<T> (const auto_delete_vec<T> &) = delete;
};
which vec.h uses is invalid C++20, one needs to use
auto_vec_delete (const auto_delete_vec &) = delete;
instead which is valid all the way back to C++11 (and without = delete
to C++98).
2020-12-02 Scott Snyder <sss@li-snyder.org>
PR plugins/98059
* vec.h (auto_delete_vec): Use
DISABLE_COPY_AND_ASSIGN(auto_delete_vec) instead of
DISABLE_COPY_AND_ASSIGN(auto_delete_vec<T>) to make it valid C++20
after DR2237.
Nathan Sidwell [Wed, 2 Dec 2020 13:52:14 +0000 (05:52 -0800)]
C++ Module Binding Vector
This adds the vector necessary to hold different module's namespace
bindings. We add a new tree-node 'tree_binding_vec', which contains a
sparse array, indexed by module number. To avoid space wasting, this
is allocated in clusters using 'unsigned short' as the index value (so
that's one of the upper bounds on module importing). If there are
only bindings from the current TU, there is no vector, so we have the
same representation as a non-module compilation.
To support lazy loading, a binding slot can contain either a tree (the
binding), or a cookie that the module machinery uses to load the
required binding on demand.
The first 2 or 3 slots end up being reserved for fixed meanings.
There are a couple of flags we have to record on a binding, to know
whether the same declaration could appear in two different slots.
Martin Liska [Wed, 2 Dec 2020 12:01:47 +0000 (13:01 +0100)]
ipa: do not DECL_IS_MALLOC for void fns
gcc/ChangeLog:
PR ipa/98075
* cgraph.c (cgraph_node::dump): Dump decl_is_malloc flag.
* ipa-pure-const.c (propagate_malloc): Do not set malloc
attribute for void functions.
which specifies the symbol name which the section references. Assembler
creates a unique __patchable_function_entries section with the section,
where foo is defined, as its linked-to section. Linker keeps a section
if its linked-to section is kept during garbage collection.
This patch checks assembler support for the section flag 'o' and uses
it to implement __patchable_function_entries section. Since Solaris may
use GNU assembler with Solairs ld. Even if GNU assembler supports the
section flag 'o', it doesn't mean that Solairs ld supports it. This
feature is disabled for Solairs targets.
gcc/
PR middle-end/93195
PR middle-end/93197
* configure.ac (HAVE_GAS_SECTION_LINK_ORDER): New. Define 1 if
the assembler supports the section flag 'o' for specifying
section with link-order.
* output.h (SECTION_LINK_ORDER): New. Defined to 0x8000000.
(SECTION_MACH_DEP): Changed from 0x8000000 to 0x10000000.
* targhooks.c (default_print_patchable_function_entry): Pass
SECTION_LINK_ORDER to switch_to_section if the section flag 'o'
works. Pass current_function_decl to switch_to_section.
* varasm.c (default_elf_asm_named_section): Use 'o' flag for
SECTION_LINK_ORDER if assembler supports it.
* config.in: Regenerated.
* configure: Likewise.
* doc/sourcebuild.texi: Document o_flag_in_section.
gcc/testsuite/
PR middle-end/93195
* g++.dg/pr93195a.C: New test.
* g++.dg/pr93195b.C: Likewise.
* lib/target-supports.exp
(check_effective_target_o_flag_in_section): New proc.
Jonathan Wakely [Wed, 2 Dec 2020 12:29:00 +0000 (12:29 +0000)]
libstdc++: Fix null pointer dereferences in __gnu_cxx::rope
This fixes UBsan errors like:
/usr/include/c++/10/ext/ropeimpl.h:593:9: runtime error: member access within null pointer of type 'struct _RopeRep'
/usr/include/c++/10/ext/ropeimpl.h:593:9: runtime error: member call on null pointer of type 'struct _Rope_rep_base'
/usr/include/c++/10/ext/rope:556:17: runtime error: reference binding to null pointer of type 'struct allocator_type'
/usr/include/c++/10/ext/ropeimpl.h:593:9: runtime error: reference binding to null pointer of type 'struct allocator_type'
/usr/include/c++/10/ext/rope:1700:30: runtime error: member call on null pointer of type 'struct new_allocator'
/usr/include/c++/10/ext/new_allocator.h:105:29: runtime error: member call on null pointer of type 'struct new_allocator'
/usr/include/c++/10/ext/rope:1702:26: runtime error: reference binding to null pointer of type 'const struct allocator'
/usr/include/c++/10/bits/allocator.h:148:34: runtime error: reference binding to null pointer of type 'const struct new_allocator'
/usr/include/c++/10/ext/rope:1664:39: runtime error: reference binding to null pointer of type 'const struct allocator'
/usr/include/c++/10/ext/rope:1665:9: runtime error: reference binding to null pointer of type 'const struct allocator_type'
/usr/include/c++/10/ext/rope:725:36: runtime error: reference binding to null pointer of type 'const struct allocator_type'
/usr/include/c++/10/ext/rope:614:64: runtime error: reference binding to null pointer of type 'const struct allocator_type'
The problem is calling r->_M_get_allocator() when r is null.
libstdc++-v3/ChangeLog:
* include/ext/rope (rope::_S_concat_char_iter)
(rope::_S_destr_concat_char_iter): Add allocator parameter.
(rope::push_back, rope::append, rope::insert, operator+):
Pass allocator.
* include/ext/ropeimpl.h (rope::_S_concat_char_iter)
(rope::_S_destr_concat_char_iter): Add allocator parameter
and use it.
(_Rope_char_ref_proxy::operator=(_CharT)): Pass allocator.
Richard Biener [Wed, 2 Dec 2020 10:07:43 +0000 (11:07 +0100)]
guard maybe_set_vectorized_backedge_value calls
This makes sure to not call maybe_set_vectorized_backedge_value when
we did not vectorize the latch def candidate.
2020-12-02 Richard Biener <rguenther@suse.de>
* tree-vect-loop.c (vect_transform_loop_stmt): Return whether
we vectorized a stmt.
(vect_transform_loop): Only call maybe_set_vectorized_backedge_value
when we vectorized the stmt.
Jakub Jelinek [Wed, 2 Dec 2020 10:33:33 +0000 (11:33 +0100)]
expansion: Fix up infinite recursion due to double-word modulo optimization
Jeff has reported that my earlier patch broke rl78-elf, e.g. with
unsigned short foo (unsigned short x) { return x % 7; }
when compiled with -O2 -mg14. The problem is that rl78 is a BITS_PER_WORD
== 8 target which doesn't have 8-bit modulo or divmod optab, but has instead
16-bit divmod, so my patch attempted to optimize it, then called
expand_divmod to do 8-bit modulo and that in turn tried to do 16-bit modulo
again.
The following patch fixes it in two ways.
One is to not perform the optimization when we have {u,s}divmod_optab
handler for the double-word mode, in that case it is IMHO better to just
do whatever we used to do before. This alone should fix the infinite
recursion. But I'd be afraid some other target might have similar problem
and might not have a divmod pattern, but only say a library call.
So the patch also introduces a methods argument to expand_divmod such that
normally we allow everything that was allowed before (using libcalls and
widening), but when called from these expand_doubleword*mod routines we
restrict it to no widening and no libcalls.
2020-12-02 Jakub Jelinek <jakub@redhat.com>
* expmed.h (expand_divmod): Only declare if GCC_OPTABS_H is defined.
Add enum optabs_method argument defaulted to OPTAB_LIB_WIDEN.
* expmed.c: Include expmed.h after optabs.h.
(expand_divmod): Add methods argument, if it is not OPTAB_{,LIB_}WIDEN,
don't choose a wider mode, and pass it to other calls instead of
hardcoded OPTAB_LIB_WIDEN. Avoid emitting libcalls if not
OPTAB_LIB or OPTAB_LIB_WIDEN.
* optabs.c: Include expmed.h after optabs.h.
(expand_doubleword_mod, expand_doubleword_divmod): Pass OPTAB_DIRECT
as last argument to expand_divmod.
(expand_binop): Punt if {s,u}divmod_optab has handler for double-word
int_mode.
* expr.c: Include expmed.h after optabs.h.
* explow.c: Include expmed.h after optabs.h.
Jakub Jelinek [Wed, 2 Dec 2020 10:32:19 +0000 (11:32 +0100)]
expansion: Further improve double-word modulo, division and divmod [PR97459]
The following patch implements what Thomas wrote about, in particular
that we can handle also double-word divison by the constants for which
the earlier patch optimized modulo (if it would be otherwise a library
call) and that we can also easily handle such constants shifted to the left.
Unfortunately, seems CSE isn't able to optimize away the two almost
identical sequences (one to compute remainder, one to compute quotient),
probably because of the ADD_OVERFLOW introduced jumps, so the patch also
adjusts expand_DIVMOD.
2020-12-02 Jakub Jelinek <jakub@redhat.com>
PR rtl-optimization/97459
* optabs.h (expand_doubleword_divmod): Declare.
* optabs.c (expand_doubleword_divmod): New function.
(expand_binop): Use it.
* internal-fn.c (expand_DIVMOD): Likewise.
* gcc.target/i386/pr97282.c (foo): Use 123456 divisor instead of
10.
* gcc.dg/pr97459-1.c (TESTS): Add tests for 10, 12 and
6144.
* gcc.dg/pr97459-2.c (TESTS): Likewise.
* gcc.dg/pr97459-3.c: New test.
* gcc.dg/pr97459-4.c: New test.
* gcc.dg/pr97459-5.c: New test.
* gcc.dg/pr97459-6.c: New test.
Kewen Lin [Wed, 2 Dec 2020 07:55:34 +0000 (01:55 -0600)]
rs6000: Disable HTM for Power10 and later by default
Power ISA 3.1 has dropped transactional memory support, this patch
is to disable HTM feature for power10 and later by default.
Bootstrapped/regtested on powerpc64le-linux-gnu P8 and P10.
gcc/ChangeLog:
* config/rs6000/rs6000.c (rs6000_option_override_internal):
Use OPTION_MASK_DIRECT_MOVE for Power8 target_enable instead
of OPTION_MASK_HTM.
* config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER):
Remove OPTION_MASK_HTM.
(RS6000_CPU): Add OPTION_MASK_HTM to power8, power9 and
powerpc64le entries.
Martin Liska [Tue, 1 Dec 2020 16:33:35 +0000 (17:33 +0100)]
diagnostics: ignore -fmax-errors for ICE
Right now I see:
./xgcc -B. ~/Programming/testcases/json.i -c -O2 -fmax-errors=1
/home/marxin/Programming/testcases/json.i: In function ‘json_variant_type’:
/home/marxin/Programming/testcases/json.i:22:1: error: non-integral type switch statement
22 | }
| ^
int *
switch (v_2(D)) <default: <L16> [INV], case 0B: <L11> [INV], case 5B: <L12> [INV], case 6B: <L13> [INV], case 7B: <L14> [INV], case 8B: <L15> [INV]>
compilation terminated due to -fmax-errors=1.
with the patch I get:
./xgcc -B. ~/Programming/testcases/json.i -c -O2 -fmax-errors=1 -c
/home/marxin/Programming/testcases/json.i: In function ‘json_variant_type’:
/home/marxin/Programming/testcases/json.i:22:1: error: non-integral type switch statement
22 | }
| ^
int *
switch (v_2(D)) <default: <L16> [INV], case 0B: <L11> [INV], case 5B: <L12> [INV], case 6B: <L13> [INV], case 7B: <L14> [INV], case 8B: <L15> [INV]>
during GIMPLE pass: iftoswitch
/home/marxin/Programming/testcases/json.i:22:1: internal compiler error: verify_gimple failed
0xe4478c verify_gimple_in_cfg(function*, bool)
/home/marxin/Programming/gcc/gcc/tree-cfg.c:5467
0xd201cf execute_function_todo
/home/marxin/Programming/gcc/gcc/passes.c:2042
0xd2101c do_per_function
/home/marxin/Programming/gcc/gcc/passes.c:1687
0xd2101c execute_todo
/home/marxin/Programming/gcc/gcc/passes.c:2096
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
gcc/ChangeLog:
* diagnostic.c (diagnostic_report_diagnostic): ICE causes to
terminate compiler immediately, so I guess it should be printed
always.
Martin Liska [Tue, 1 Dec 2020 11:18:46 +0000 (12:18 +0100)]
if-to-switch: Support chain with 2 BBs.
As seen in the test-case, even 2 BBs can handle interesting
cases covered by a jump table or a bit-test.
gcc/ChangeLog:
PR tree-optimization/88702
* gimple-if-to-switch.cc (pass_if_to_switch::execute):
Require at least 2 BBs.
* gimple-if-to-switch.cc (find_conditions): Require
equal precision for low and high of a range.
gcc/testsuite/ChangeLog:
PR tree-optimization/88702
* gcc.dg/tree-ssa/if-to-switch-9.c: New test.
Jeff Law [Wed, 2 Dec 2020 04:48:10 +0000 (21:48 -0700)]
Use add/sub/neg insns to eliminate compare/test insns on H8
gcc/
* config/h8300/addsub.md (addqi3_clobber_flags): Rename to
addqi3_flags and annotate with a <cczn> for define_subst.
(addhi3_h8sx_clobber_flags): Likewise.
(subqi3_clobber_flags, sub<mode>3_clobber_flags): Likewise.
(neg<mode2>_clobber_flags): Similarly.
(addsi3_clobber_flags): Similarly. Update last argument to
output_plussi to distinguish when we need flags or do not need
flags.
(addhi3_clobber_flags): Similarly. Twiddle code for cases
+-1, +-2 and +-4.
* config/h8300/h8300.md: Define iterators, mode attributes and
substitutions for use in compare/test elimination.
* config/h8300/jumpcall.md (branch, branch_1): Use H8cc mode
iterator to cover the different modes for the CC register.
(branch_1_false): Likewise.
gcc/testsuite
* gcc.target/h8300/add.c: New test.
* gcc.target/h8300/add-2.c: New test.
* gcc.target/h8300/add-3.c: New test.
* gcc.target/h8300/sub.c: New test.
* gcc.target/h8300/sub-2.c: New test.
* gcc.target/h8300/sub-3.c: New test.
H.J. Lu [Mon, 3 Feb 2020 19:55:43 +0000 (11:55 -0800)]
Use SHF_GNU_RETAIN to preserve symbol definitions
In assemly code, the section flag 'R' sets the SHF_GNU_RETAIN flag to
indicate that the section must be preserved by the linker.
Add SECTION_RETAIN to indicate a section should be retained by the linker
and set SECTION_RETAIN on section for the preserved symbol if assembler
supports SHF_GNU_RETAIN. All retained symbols are placed in separate
sections with
which places not_preserved_symbol definition in the SHF_GNU_RETAIN
section.
gcc/
2020-12-01 H.J. Lu <hjl.tools@gmail.com>
* configure.ac (HAVE_GAS_SHF_GNU_RETAIN): New. Define 1 if
the assembler supports marking sections with SHF_GNU_RETAIN flag.
* output.h (SECTION_RETAIN): New. Defined as 0x4000000.
(SECTION_MACH_DEP): Changed from 0x4000000 to 0x8000000.
(default_unique_section): Add a bool argument.
* varasm.c (get_section): Set SECTION_RETAIN for the preserved
symbol with HAVE_GAS_SHF_GNU_RETAIN.
(resolve_unique_section): Used named section for the preserved
symbol if assembler supports SHF_GNU_RETAIN.
(get_variable_section): Handle the preserved common symbol with
HAVE_GAS_SHF_GNU_RETAIN.
(default_elf_asm_named_section): Require the full declaration and
use the 'R' flag for SECTION_RETAIN.
* config.in: Regenerated.
* configure: Likewise.
* doc/sourcebuild.texi: Document R_flag_in_section.
gcc/testsuite/
2020-12-01 H.J. Lu <hjl.tools@gmail.com>
Jozef Lawrynowicz <jozef.l@mittosystems.com>
* c-c++-common/attr-used.c: Check the 'R' flag.
* c-c++-common/attr-used-2.c: Likewise.
* c-c++-common/attr-used-3.c: New test.
* c-c++-common/attr-used-4.c: Likewise.
* gcc.c-torture/compile/attr-used-retain-1.c: Likewise.
* gcc.c-torture/compile/attr-used-retain-2.c: Likewise.
* lib/target-supports.exp
(check_effective_target_R_flag_in_section): New proc.